Transcript r - Earth & Planetary Sciences

EART162: PLANETARY
INTERIORS
Francis Nimmo
F.Nimmo EART162 Spring 10
Last Week
• Elasticity: Young’s modulus = stress / strain (also
Poisson’s ratio – what does it do?)
• Another important (why?) variable is the bulk modulus,
which tells us how much pressure is required to cause a
given change in density. Definition?
• Flow law describes the relationship between stress and
strain rate for geological materials
• (Effective) viscosity is stress / strain rate
• Viscosity is very temperature-dependent: exp(-Q/RT)
F.Nimmo EART162 Spring 10
This Week – Isostasy and Flexure
•
•
•
•
•
•
See Turcotte and Schubert chapter 3
How are loads supported?
Isostasy – zero elastic strength
Flexure – elastic strength (rigidity) is important
What controls rigidity?
We can measure rigidity remotely, and it tells us
about a planet’s thermal structure
F.Nimmo EART162 Spring 10
Airy Isostasy
• Let’s start by assuming that the crust and mantle are
unable to support loads elastically
Load
Load
Crust rc
Mantle rm
• The crust will deflect downwards until the surface
load (mass excess) is balanced by a subsurface “root”
(mass deficit – dense mantle replaced by light crust)
• 90% of an iceberg is beneath the surface of the ocean
for exactly the same reason
• This situation is called (Airy) isostasy
F.Nimmo EART162 Spring 10
Consequences of Isostasy
h
Crust rc
Constant
pressure
r
tc
r
Mantle rm
• In the case of no elastic strength,
the load is balanced by the
mantle root: hrc=r(rm-rc)
• This also means that there are
no lateral variations in pressure
beneath the crustal root
• So crustal thickness contrasts (Dtc=h+r) lead to
elevation contrasts (h):
(r  r )
h
m
rm
c
Dtc
• Note that the elevation is independent of the
background crustal thickness tc
F.Nimmo EART162 Spring 10
Pratt Isostasy
• Similar principle – pressures below some depth do
not vary laterally
• Here we do it due to variations in density, rather than
crustal thickness
r2 > r1
tc
r1
Mantle rm
h
r2
h
( r 2  r1 )
r2
tc
• What’s an example of where this mechanism occurs
on Earth?
F.Nimmo EART162 Spring 10
Gravity Effects
• Because there are no lateral variations in pressure
beneath a certain depth, that means that the total mass
above this depth does not vary laterally either
• So what sort of gravity anomalies should we see?
(NB there will actually be a
• Very small ones!
small gravity anomaly and
gravity
gravity
edge effects in this case)
Crust rc
Uncompensated load – Dg=2prcG
Compensated load – Dg~0
• So we can use the size of the gravity anomalies to tell
whether or not surface loads are compensated
F.Nimmo EART162 Spring 10
Example - Mars
• The southern half of Mars is about 3 km higher than the
northern half (why?)
• But there is almost no gravity anomaly associated with this
“hemispheric dichotomy”
• We conclude that the crust
of Mars here must be
compensated (i.e. weak)
• Pratt isostasy? Say
r1=2700 kgm-3 (granite)
and r2=2900 kgm-3
(basalt). This gives us a
crustal thickness of 45 km
F.Nimmo EART162 Spring 10
Mars (cont’d)
• On the other hand, some of the big
volcanoes (24 km high) have gravity
anomalies of 2000-3000 mGal
• If the volcanoes were sitting on a
completely rigid plate, we would
expect a gravity anomaly of say 2.9 x
24 x 42  2900 mGal
• We conclude that the Martian
volcanoes are almost uncompensated,
so the crust here is very rigid
Olympus
Ascraeus
Pavonis
Arsia
Crust rc
• Remember that what’s important is the
strength of the crust at the time the load
was emplaced – this may explain why
different areas have different strengths Uncompensated load – Dg=2prcG
F.Nimmo EART162 Spring 10
Flexure
• So far we have dealt with two end-member cases: when
the lithosphere is completely rigid, and when it has no
strength at all (isostasy)
• It would obviously be nice to be able to deal with
intermediate cases, when a load is only partly supported
by the rigidity of the lithosphere
• I’m not going to derive the key equation – see the
Supplementary Section (and T&S Section 3-9) for the
gory details
• We will see that surface observations of deformation can
be used to infer the rigidity of the lithosphere
• Measuring the rigidity is useful because it is controlled
by the thermal structure of the subsurface F.Nimmo EART162 Spring 10
Flexural Stresses
load
Crust
Elastic plate
Mantle
• In general, a load will be supported by a combination of
elastic stresses and buoyancy forces (due to the different
density of crust and mantle)
• The elastic stresses will be both compressional and
extensional (see diagram)
• Note that in this example the elastic portion includes both
crust and mantle
F.Nimmo EART162 Spring 10
Flexural Equation (1)
q(x)
w(x) rw
rc
Crust
Te
rm
Mantle
Elastic plate
P
P
P is force per unit length
in the z direction
4
2
d w
d w
D 4  P 2  r m  r w gw  q( x)
dx
dx
• D is the (flexural) rigidity, Te is the elastic thickness
ETe3
D
2
12(1  v )
F.Nimmo EART162 Spring 10
Flexural Equation (2)
h
rc
w
rm
rm
w
rl
Te
• Here the load q(x)=rlgh
• We’ll also set P=0
• The flexural equation is:
d 4w
D 4  r m  rl gw  rl gh
dx
• If the plate has no rigidity, D=0 and we get
( rm  rl )w  hrl
• This is just the expression for Airy isostasy
• So if the flexural rigidity is zero, we get isostasy
F.Nimmo EART162 Spring 10
Flexural Equation (3)
d 4w
D 4  r m  rl gw  rl gh
dx
What’s this?
• Let’s assume a sinusoidal variation in loading h=h0eikx
• Then the response must also be sinusoidal w=w0eikx
• We can relate h0 to w0 as follows
a
w0 
rl
Dr 
Dk 4
g
h0
Here Drrm-rl and k=2p/l,
where l is the wavelength
• Does this expression make sense?
• What happens if D=0 or Dr=0?
• What happens at very short or very long wavelengths?
F.Nimmo EART162 Spring 10
Degree of Compensation
• The deflection caused by a given load: w0 
rl
Dr 
• We also know the deflection in the case of a
rl
completely compensated load (D=0): w1  Dr h0
Dk 4
g
h0
• The degree of compensation 1 Short l:
C Uncompensated
C is the ratio of the
deflection to the deflection 0.5
Long l:
in the compensated case:
Compensated
0
w0
1
C

Dk4/Drg=1 wavelength
Dk 4
w1 1  gDr
• Long l, C~1 (compensated); short l, C~0 (uncomp.)
• C~1 gives small gravity anomalies, C~0 large ones
1/4
• Critical wavenumber: C=0.5 means k=(Drg/D)
F.Nimmo EART162 Spring 10
Example
• Let’s say the elastic thickness on Venus is 30 km
(we’ll use E=100 GPa, v=0.25, g=8.9 ms-2, Dr=500 kg m-3)
• The rigidity D=ETe3/12(1-v2) ~ 2x1023 Nm
• The critical wavenumber k=(Drg/D)1/4 ~1.3x10-5 m-1
• So the critical wavelength l=2p/k=500 km
1.6 km
0.8 km
1000 km
2000 km
3000 km
Would we expect it to be compensated or not?
What kind of gravity anomaly would we expect?
F.Nimmo EART162 Spring 10
Degree of Compensation (2)
~0 mGal/km
1
C
0.5
Weaker (small Te)
Short l:
Uncompensated
Stronger (large Te)
Long l:
Compensated
~120 mGal/km 0
wavelength
• What gravity signals are associated with C=1 and C=0?
• How would the curves move as Te changes?
So by measuring the ratio of gravity to topography
(admittance) as a function of wavelength, we can infer
the elastic thickness of the lithosphere remotely
F.Nimmo EART162 Spring 10
Flexural Parameter (1)
rw
• Consider a line load
acting on a plate:
load
w(x)
w0
Te
x=0
rm
x
• Except at x=0, load=0 so we can write:
What does this
d 4w
D 4  r m  r w gw  0 look like?
dx
• Boundary conditions for an unbroken plate
are that dw/dx=0 at x=0 and w  0 as x  
• The solution is w  w0 exp( x / a )cos ax  sin ax 
• Here a is the flexural parameter
a
(Note the similarity of a to the critical wavenumber)

4D
g ( rm  rw )
a

1
4
F.Nimmo EART162 Spring 10
Flexural Parameter (2)
• Flexural parameter a=(4D/g(rm-rw))1/4
• It is telling us the natural wavelength of the elastic plate
• E.g. if we apply a concentrated load, the wavelength of
the deformation will be given by a
• Large D gives long-wavelength deformation, and v.v.
• If the load wavelength is >> a, then the deformation will
approach the isostatic limit (i.e. C~1)
• If the load wavelength is << a, then the deformation will
be small (C~0) and have a wavelength given by a
• If we can measure a flexural wavelength, that allows us
to infer a and thus D or Te directly. This is useful!
F.Nimmo EART162 Spring 10
10 km
Distance, km
•
•
•
•
Example
• This is an example of a profile
across a rift on Ganymede
• An eyeball estimate of a
would be about 10 km
• For ice, we take E=10 GPa,
Dr=900 kg m-3 (there is no
overlying ocean), g=1.3 ms-2
If a=10 km then D=2.9x1018 Nm and Te=1.5 km
A numerical solution gives Te=1.4 km – pretty good!
So we can determine Te remotely
This is useful because Te is ultimately controlled by
the temperature structure of the subsurface
F.Nimmo EART162 Spring 10
Te and temperature structure
• Cold materials behave elastically
• Warm materials flow in a viscous fashion
• This means there is a characteristic temperature
(roughly 70% of the melting temperature) which
defines the base of the elastic layer
Depth
•E.g. for ice the base of the elastic
110 K
270 K
190 K
layer is at about 190 K
1.4 km
• The measured elastic layer thickness
elastic
is 1.4 km (from previous slide)
• So the thermal gradient is 60 K/km
viscous
• This tells us that the (conductive) ice
Temperature
shell thickness is 2.7 km (!)
F.Nimmo EART162 Spring 10
Te and age
McGovern et al., JGR 2002
Small Te
Decreasing age
• The elastic thickness
recorded is the lowest
since the episode of
deformation
• In general, elastic
thicknesses get larger
with time (why?)
Large Te
• So by looking at features of different ages, we can
potentially measure how Te, and thus the temperature
structure, have varied over time
• This is important for understanding planetary evolution
F.Nimmo EART162 Spring 10
Te in the solar system
• Remote sensing observations give us Te
• Te depends on the composition of the material (e.g. ice,
rock) and the temperature structure
• If we can measure Te, we can determine the temperature
structure (or heat flux)
• Typical (approx.) values for solar system objects:
Body
Te (km)
Earth
(cont.)
Mars
(recent)
Europa
30
dT/dz
(K/km)
15
100
5
2
40
Body
Te
Venus
30
(450oC)
Moon
15
(ancient)
Ganymede 2
dT/dz
(K/km)
15
30
40
F.Nimmo EART162 Spring 10
Summary
• Flexural equation determines how loads are supported:
d 4w
D 4  r m  r w gw  q( x)
dx
• The flexural parameter a gives us the characteristic
wavelength of deformation
a


1/ 4
4D
g ( rm rw )
• Loads with wavelengths >> a are isostatically supported
• Loads with wavelengths << a are elastically supported
• We can infer a from looking at flexural topography (or by
using gravity & topography together – admittance)
• Because the rigidity depends on the temperature structure,
determining a allows us to determine dT/dz F.Nimmo EART162 Spring 10
Supplementary Material follows
F.Nimmo EART162 Spring 10
l
Plate Bending
• Bending an elastic plate
y
Dl
produces both compressional
compression
and extensional strain
• The amount of strain
f
R
depends on the radius of
R
curvature R
Dl yf y

Strain   xx 
f
l
Rf R
• Note that there is no strain along the centre line (y=0)
• The resulting stress is given by (see T&S):
E
 xx 

2 xx
1 v
F.Nimmo EART162 Spring 10
extension
Radius of Curvature
• What is the local radius of curvature of a deforming
plate? Useful to know, because that allows us to calculate
the local stresses
It can be shown that
Dx
2
a
R
R
q2
w
q1
x
1
d w
 2
R
dx
So the bending stresses are given by
2
E d w
 xx   y
A
2
2
1  v dx
(y is the distance from the mid-plane)
Plate shape
described by w(x)
F.Nimmo EART162 Spring 10
Bending Moment
q(x) (force per unit area)
q(x)
V+dV
(shear force per
unit length)
V
M+dM
(moment)
M
dx
• Balance torques: V dx = dM
• Balance forces: q dx + dV = 0
• Put the two together:
d 2M
 q
2
dx
B
Does this make sense?
y
xx
Te/2
Moment: M 
Te / 2

Te / 2
xx
ydy C
F.Nimmo EART162 Spring 10
Putting it all together . . .
• Putting together A, B and C we end up with
a
3
e
4
4
ET
d w
d w
 D 4  q ( x)
2
4
12(1  ) dx
dx
• Here D is the rigidity
• Does this equation make sense?
F.Nimmo EART162 Spring 10
Admittance Example
Nimmo and McKenzie 1998
• Comparison of Hawaii and Ulfrun Regio (Venus)
• What is happening on Venus at short wavelengths?
• Are you surprised that the two elastic thicknesses are
comparable?
F.Nimmo EART162 Spring 10